ABB Robotics' 2026 product selector lists a complete cobot and industrial robot portfolio covering 6-axis articulated arms, mobile robots, controllers, and software built for modern shared-workspace manufacturing [S1]. On the academic side, Springer Nature's 2024 comparative study measured manual versus cobot-assisted assembly and found ergonomic and ease-of-execution gains without substantial loss of cycle time when the cobot was operated in coexistence, cooperation, and collaboration modes [S2].
Process engineers planning a cobot cell must reconcile four spec layers: the manipulator's mechanical stack (reach, payload, joint torque sensing), the safety-rated monitored stop and power-and-force-limiting (PFL) functions called up by ISO 10218-1 and ISO TS 15066, the controller's I/O and fieldbus map, and the gripper/end-effector stack sized to the part [S3][S4]. ABB frames its cobot line as one element inside a wider portfolio that also includes traditional industrial and collaborative robots, mobile platforms, and software, so cell builders should pick the arm first and the surrounding automation around it [S1].
Cell Architecture: From Coexistence to True Collaboration
The Springer 2024 benchmark defines three operating modes — coexistence (shared space, sequential work), cooperation (shared space, alternating work on the same part), and collaboration (shared space, simultaneous work on the same part) — and reports that even the lightest coexistence mode produced measurable ergonomic improvement versus a fully manual bench [S2]. The 2020 CIRP-anchored overview in the same Springer series frames HRC as requiring the robot to dynamically adapt pre-planned tasks to human behaviour, which in practice means the controller, not the operator, absorbs the variability [S3].
Cell layouts that go beyond coexistence typically use a probabilistic workspace-sharing model (Pellegrinelli et al., CIRP Ann. 2016) to keep the human and the arm out of the same swept volume above a defined separation speed, with the cobot dropping to a monitored stop when the operator enters the safeguarded space [S3]. Reference cells cited in the Springer literature run KUKA LBR iiwa arms paired with SCHUNK Co-Act EGP-C grippers, a combination that exposes joint torque sensing all the way to the end-effector and lets the PFL thresholds be set per joint rather than per cell [S2].
Manipulator Stack: Axes, Reach, Payload and Joint Sensing
Cobots ship in three mechanical templates: 6-axis articulated arms for general assembly, 7-axis arms (KUKA LBR iiwa, FANUC CRX, Universal Robots UR20 class) for reaching around fixtures, and SCARA-style 4-axis arms for high-speed pick-and-place. ABB's articulated portfolio is structured around 6-axis units with payload and reach tiers matched to the application selector on its robotics site [S1].
Payload tiers in current commercial cobots span roughly 3 kg (UR3/e-Series class), 5–8 kg (UR5, KUKA LBR iiwa 7 R800), 12–16 kg (UR10, Doosan A0509), 20–30 kg (UR20, FR20) — exact payloads vary by manufacturer and the Springer 2022 human-sensor framework paper recommends validating the working payload against the worst-case part mass plus gripper mass, not the catalogue maximum [S4]. Repeatability for cobots is typically ±0.03 mm to ±0.1 mm, one order of magnitude looser than industrial articulated robots, so they trade absolute precision for compliance and ease of redeployment [S2][S4].
Safety Stack: ISO 10218-1, ISO TS 15066 and Power-and-Force Limiting
ISO 10218-1:2011 governs safety requirements for industrial robots and is the parent document for collaborative operation, with ISO TS 15066 supplying the biomechanical thresholds (quasi-static contact, transient contact, and clamping limits) that a PFL-rated cell must prove it stays under [S3]. Chemweno, Pintelon and Decre's 2020 Safety Science review of ISO TS 15066 walks through the four collaborative operations — safety-rated monitored stop, hand guiding, speed and separation monitoring, and power-and-force limiting — and ties each one to a hazard-analysis method the integrator must document [S2].
The minimum required cell-level functions are: an emergency stop on each operator station that is hard-wired to the safety controller, a safety-rated monitored stop triggered by a light curtain or area scanner, and a PFL configuration that caps joint torque, TCP speed, and end-effector energy below the ISO TS 15066 biomechanical limits for the body region exposed to contact [S2][S3]. Buerkle et al.'s 2022 adaptive human-sensor framework paper stresses that the safety case should be re-validated when a new tool mass, a new payload, or a new operator training level is introduced, because the PFL thresholds are set against the worst-case contact scenario in the risk assessment [S4].
Controller, Fieldbus and Software Stack
Cobot controllers expose the same digital I/O blocks used by industrial arms (24 V DC digital in/out, safety-rated I/O, sometimes EtherCAT or PROFINET on the safety bus) but add a teach-pendant or hand-guiding interface that lets a non-programmer record waypoints by physically moving the arm [S1][S3]. The 2020 Springer overview notes that programming-free and multimodal communication (hand guiding, lead-through, and graphical programming on the HMI) is now a baseline expectation, replacing the rigid native-code path that older industrial robots required [S3].
ABB's product page groups the controller, software, and equipment around the robot so that cell builders source the safety scanner, gripper, and conveyor handshakes from a single vendor stack [S1]. On the open-source side, the May 2026 GitHub project "collaborative_robot_manipulator_software_system" by EmanElRify publishes a 2-DOF arm reference design with serial-comm firmware and a system architecture diagram that downstream teams can fork as a starting point, though it sits well below commercial payload and safety certifications [S5].
Assembly Economics: Cycle Time vs Ergonomic Gain
Barravecchia and Mastrogiacomo's 2023 cost-model paper in the International Journal of Advanced Manufacturing Technology frames the cobot decision as a comparison between labour cost saved by automation, the amortised capital cost of the cobot cell, and the residual ergonomic risk premium on the manual line. The Springer 2024 experimental comparison found that introducing a cobot into a manual assembly bench improved ergonomics and ease of execution without substantially compromising assembly time, which means the payback is driven by injury-rate reduction and uptime, not cycle-time compression [S2][S6].
For a process engineer sizing a new cell, the order of operations is: (1) pick the operating mode (coexistence, cooperation, collaboration) from the assembly sequence, (2) pick the arm payload and reach from the worst-case part, (3) pick the gripper stack, (4) run the ISO TS 15066 biomechanical calc to set PFL limits, and (5) wire the safety I/O to a safety PLC or integrated safety controller [S2][S3]. Greening's 2018 AHP-based cobot selection methodology in the International Journal of Rapid Manufacturing argues for scoring arms on a weighted set of criteria — payload, reach, repeatability, safety functions, programming method, vendor support — rather than buying on catalogue price alone [S7].
Failure Modes, Constraints and Selection Boundary
Cobots are the wrong tool when a process demands high-speed operation above the PFL ceiling, sub-0.05 mm repeatability, or a long duty cycle in a dirty, oily, or high-temperature cell that would overwhelm the joint torque sensors. They are the right tool for small-batch assembly, screw-driving, machine tending on low-variance parts, and any workstation where the same operator must perform a precision sub-task thousands of times per shift [S2][S3].
Key constraints captured in the literature: the PFL thresholds force a TCP speed cap of 250 mm/s during collaborative contact operations, the workspace must remain free of pinch points that could trap a finger, and the operator training programme must be documented as part of the safety file, not left to the line supervisor [S2][S3]. Research Square's 2022 Assembly System 4.0 paper also flags that assembly-area cobot deployments need a documented change-control process for fixturing, because a new fixture can move the human-robot pinch geometry outside the original ISO TS 15066 envelope [S8].
Track the IEEE/ISO working group updates to ISO TS 15066 biomechanical limit tables and the IFR World Robotics cobot install density; both are the two indices that historically move before a major cell re-spec. Also watch the additive-manufacturing material supply side, because printed end-effector brackets and jigs are increasingly integrated into cobot cells to cut fixturing lead time, and the relevant machine-vision integration stack is what lets a cobot cell see a partially fixtured part without a hard-coded fixture offset.
For component-level specifications, see multifunction process calibrator.